Yuen Long 4,200-Unit Development: Reviewing a Marginal Traffic Impact Assessment
A developer proposes 4,200 flats, a 9,000 m² shopping centre and a primary school on a 6.2 ha former industrial site in Yuen Long. The only consented access is a signalised T-junction onto a four-lane divided trunk road carrying 2,650 vehicles per hour in the AM peak. The proponent's Traffic Impact Assessment uses a deterministic junction model, reports a degree of saturation of 0.99 for the 2031 design year, and concludes the impact is "acceptable with mitigation", the mitigation being signal re-timing. The 95th-percentile queue at the junction is 380 m; 240 m upstream is a roundabout that also serves a bus terminus. The site is 700 m from a planned intermediate station on the running line, due to open in 2032. Parking is proposed at 1.1 spaces per flat. The Transport Department has returned the TIA for review before resubmission.
1. As reviewing engineer, set out the deficiencies in the TIA and the specific modelling assumptions you would challenge before resubmission.
Read the worked answer
The headline number condemns the scheme, not saves it. A degree of saturation of 0.99 means the junction is at practical capacity with essentially no reserve; the report should be quoting practical reserve capacity and demonstrating headroom under a design-year growth sensitivity, not calling 0.99 acceptable.
Model class is wrong for the symptom. Deterministic junction models truncate queues and cannot represent blocking. A 380 m 95th-percentile queue against a roundabout 240 m upstream is a blocking-back problem: once the exit is blocked, the roundabout locks and the bus terminus with it. That requires microsimulation, calibrated and validated to observed flows (GEH under 5 on counts and queues), with signal coordination and pedestrian stages represented.
Trip generation and internalisation. Demand should be derived from the Hong Kong Planning Standards and Guidelines Chapter 8 rates, with explicit internalisation, pass-by and diverted-trip assumptions for the retail, and a school peak at about 07:45 that is offset from the commuter peak but not from the bus terminus. The report should show the trip matrix, not a single cordon total.
Missing sensitivities. No 2031 growth case without the development; no case with the station open (feeder buses, drop-off, pedestrian demand, and the construction interface if the station works overlap); no assessment of the 1.1 spaces per flat against prevailing parking policy, which suppresses or inflates demand depending on how it is treated; no construction-phase traffic management; no pedestrian, cyclist or school-safety audit.
Any one of these alone is arguable; together they mean the 0.99 figure is not a validated result and the "acceptable" conclusion has no basis yet.
2. Recommend a mitigation package and justify its sequencing against the 2032 station opening.
Read the worked answer
Physical works (before first occupation). A dedicated left-turn slip lane and a lengthened right-turn pocket raise practical capacity cheaply; a second access onto the district distributor removes single-point dependence and is the single most valuable item, but it requires about 900 m² of government land and a road reserve gazettal, so it must start immediately to be available at occupation. Signal control should be coordinated with the roundabout, or replaced by an adaptive controller with queue detectors on the northbound approach, with a hard rule that the junction must not discharge into a blocked roundabout.
Demand-side measures (at occupation). Cap parking at 0.4–0.5 spaces per flat with unbundled parking, a car club of 8–10 vehicles, and a travel plan with a funded coordinator, a shuttle to the existing town centre and subsidised monthly transit passes for the first three years. At 4,200 flats this is not decoration; parking supply is the most elastic lever available and is far cheaper than a grade separation.
Sequencing around 2032. The station changes the trip profile but not in time to rescue a 2031 design year. So: interim package (slip lane, second access, parking cap, travel plan) delivered before occupation; a monitoring and review gate at 60% occupation; then a station-integration package (bus-rail interchange, walking and cycling links, removal of temporary parking, re-timing of the shuttle) built with the station. Do not fund the permanent works on the assumption of station-mode shift - model it as a sensitivity, because the 700 m walk is attractive but the feeder bus service is not yet committed.
3. What residual risk would you accept, and what monitoring and enforcement would you attach to the approval?
Read the worked answer
Residual risk. Even with the package, the junction sits at a high degree of saturation in the AM peak and the roundabout remains sensitive to any incident or bus breakdown at the terminus. I would accept that as an operational risk, not a safety risk, provided the spillback protection is physical or enforced by detection rather than assumed.
Monitoring. A condition requiring continuous traffic and queue monitoring at the junction and roundabout for the first three years, with quarterly reporting against the predicted flows, a defined trigger (sustained 95th-percentile queue beyond 300 m, or roundabout blocking on more than 5% of peak intervals) and a pre-agreed remediation menu with costed options. A developer-funded travel plan review at years 1, 3 and 5, with mode-share targets, not just deliverables.
Enforcement. A bond or financial guarantee sized to the second-access and signal works, released only on commissioning evidence; a restriction on the occupation of phases beyond a stated unit count until the second access is open; and a car-park management plan that cannot be varied without Transport Department consent, otherwise the parking cap is the first thing sold off.
Professional position. Record the residual risk and the basis of acceptance in writing in the review report. An approval that quietly converts a 0.99 saturation into "acceptable with mitigation" without a monitoring obligation transfers the risk to future residents and to the department, which is not a position a professional engineer should sign.